1

Gender Gaps in Performance: Evidence from Young Lawyers

Ghazala Azmat† Rosa Ferrer‡

March 2012

Abstract

This paper documents and studies the gender gap in performance among associate

lawyers in the United States. Unlike most high-skilled professions, the legal profession

has widely-used objective methods to measure and reward lawyers’ productivity: the

number of hours billed to clients and the amount of new-client revenue generated. We

find clear evidence of a gender gap in annual performance with respect to both

measures. Male lawyers bill ten-percent more hours and bring in more than double the

new-client revenue. We show that the differential impact across genders in the presence

of young children and the differences in aspirations to become a law-firm partner

account for a large part of the difference in performance. These performance gaps have

important consequences for gender gaps in earnings. While individual and firm

characteristics explain up to 50 percent of earnings gap, the inclusion of performance

measures explains most of the remainder.

Keywords: performance measures, gender gaps, lawyers JEL classification: M52, J16, K40, J44

The views and conclusions stated herein are ours and do not necessarily reflect the views of individuals or organizations associated with the “After the JD Study.” Financial support by the Spanish Commission of Science and Technology (ECO2011-30323-C03-02, ECO2010-15052 and ECO2008-01116), the Barcelona GSE research network, and the Government of Catalonia is gratefully acknowledged. † Universitat Pompeu Fabra, Barcelona GSE, Queen Mary University, and CEP (LSE). Email:ghazala.azmat@upf.edu ‡ Universitat Pompeu Fabra and Barcelona GSE. Email:rosa.ferrer@upf.edu

2

1. Introduction

The reasons behind the gender gaps in career outcomes, particularly among

high-skilled workers, remain unclear. Due to the complexity of measuring performance,

it is difficult to understand what part—if any—of these gaps can be attributed to

differences in performance. Firms reward higher individual performance either directly,

through performance pay, or indirectly, through promotion and hiring decisions.

Therefore, to analyze differences in career outcomes, it is important to determine

whether there exist gender differences in performance and what could be driving them.

This is particularly true for high-skilled workers since a growing number of them are

explicitly evaluated and compensated based on performance.

In this paper, we document gender differences in performance among high-

skilled workers. Performance indicators are often costly to gather and heterogeneous

across industries and firms. This poses difficulties in measuring performance

differences across workers and evaluating their implications. We overcome this problem

by using the legal profession as a setting. Unlike many other sectors, the legal

profession traditionally uses performance measures that are transparent and

homogeneous across firms. In our analysis, we use comprehensive information on

nationally representative young lawyers in the U.S., including detailed information on

the measures by which they are evaluated: annual hours billed and the amount of new-

client revenue brought to the firm.1 These measures have become widely used within

the profession in the last several decades, as a means not only to explicitly compensate

lawyers, but also to evaluate lawyers for promotion decisions.

We find substantial gender differences in annual performance. To understand

the performance gap, we explore the more traditional explanations of discrimination,

child-rearing, and human-capital differences. We also consider differences in

preferences regarding areas of specialization, the inclination toward overbilling,

networking behavior, and career aspirations. We find that while the presence of pre-

school children in the household has a crucial differential effect on the performance of

male and female lawyers, differences in aspirations to “become a partner” in the law

firm is also a key determinant of the gap. In particular, such aspirations affect the

amount of new-client revenue, the performance measure that is more relevant for long-

1 Private-practice lawyers record the amount of time they devote to each case, and, based on these records, law firms determine the value of the legal service provided and clients’ fees. Moreover, most law firms assess performance and determine lawyers’ bonus based on these records (Fox, 2007).

3

run career outcomes. In contrast, other preference-related explanations that we consider

are less relevant in explaining gender gaps in performance. For instance, female lawyers

have a lower tendency to report “overbilling” of clients, and, although overbilling has

(positive) consequences in terms of performance, it has a negligible effect in explaining

gender gaps. In a similar way, the amount of time spent networking, which differs

significantly between male and female lawyers, does not explain a substantial part of the

gender differences in the performance. With regards to discrimination, it is possible that

the key determinants of the performance differences — child rearing and career

aspirations— are associated to subtle forms of discrimination, such as compliance with

social norms. However, a key finding of the paper is that the gender gaps in

performance do not appear to be caused by explicit discrimination at the firm level.

Performance differences are often a key determinant of career advancement.

Traditionally, researchers have relied on proxies for performance, such as differences in

absenteeism (Ichino and Moretti, 2010), to understand gender gaps in labor-market

outcomes. However, here we can use on-the-job performance to understand gender gaps

in career profiles. We explore the implications of these differences by focusing on

consequences in terms of earnings. As in other professions and industries, the legal

profession has a persistent gender gap in earnings. Figure 1 illustrates the large gender

difference in lawyers’ median salary; moreover, it shows that there is no evidence of

this difference decreasing in the last decade as the male-dominated generations retire.2

We find that the gap in performance helps to explain lawyers' gender gap in

earnings, a considerable proportion of which has remained unexplained until now

(Wood et al., 1993; Dinovitzer et al., 2009). This has been the case more generally in

the gender literature, in which, despite the wide range of explanatory measures, a large

part of the gender gap in earnings is unaccounted for (see Altonji and Blank, 1999, for a

review of the literature). We find a raw gender earnings gap between male and female

lawyers of 25 log points. When we control for individual and firm characteristics, we

can explain 50 percent of this initial gap. Controlling for performance, we are able to

explain almost all of the remaining gender gap. Our paper highlights that there exists an

omitted-variable bias problem since to proxy for performance—which is commonly

omitted—it is not sufficient to simply control for the observable characteristics typically

used in the literature. The previous literature has explored various possible sources of

2 It was not until the 1980s that the expansion of the legal profession attracted a large number of women (Rosen, 1992).

4

the gap, but their relative importance and whether they are related—not only to

earnings, but also to performance—remains unclear.

Recent research has explored the importance of performance pay on inequality

across economic sectors.3 The link between performance pay and increased inequality is

found to be stronger for high-income individuals because, for them, salaries may be

more sensitive to individual productive characteristics rather than job-specific

characteristics (Lemieux et al., 2009). Thus, focusing on professions in the top part of

the income distribution can provide further insights into these findings.

The legal profession is among the highest-paid professions in the U.S., along

with physicians and CEOs,4 and it constitutes a substantial part of the U.S. GDP.5 There

has been increased interest in why large earning gaps exist among the more able and

more career-driven women in high-skilled professions (Manning and Swaffield, 2008;

Bertrand et al., 2010).6 Attending law school is costly in terms of fees, opportunity

costs, and living expenses.7 Moreover, the high-skilled nature of the profession suggests

that male and female lawyers have similar skills, training and motivation. In particular,

we focus on a generation of lawyers that experienced virtual gender equality in law

school admissions and no prominent gender differences in law school performance.

Our paper documents large gender gaps in performance, and we show that these

gaps have important subsequent consequences on earnings. Because we use detailed

data on typically omitted variables, such as performance, our findings contribute to

3 Lemieux et al. (2009) study the evolution of performance pay and wage inequality in the U.S. labor- market system from the 1970s to the 1990s. Heywood and Parent (2009) use the same period but focus on the white-black wage gap. They find that the white-black earnings differential is larger in the share of the income distribution where performance pay is more prevalent. Finally, comparing Spanish industries, De la Rica et al. (2010) find that the gender gap is considerably larger for workers whose salaries include a variable component compared to those who have a fixed salary. 4 National Cross-Industry wage estimates, U.S. Bureau of Labor Statistics. 5In 2008, legal expenses accounted for more than $200 billion, which constituted 1.5 percent of the U.S. GDP. Comparing it to other large economic sectors, we observe that this was $80 billion more than educational services and almost four times more than air transportation services (Bureau of Economic Analysis, U.S. Department of Commerce). 6 Manning and Swaffield (2008) and Bertrand et al. (2010) both find that there is no gender earning gap in the outset of young professionals’ careers, but that their earnings diverge ten years after graduation. Bertrand et al. (2010) focus on MBA graduates from the University of Chicago and attribute growing earning-gap differences to career disruptions; training choices prior to MBA graduation; and weekly hours worked. Manning and Swaffield (2008) focus on graduates in the UK and find that differences in human capital and psychological factors explain part of the wage-growth gap, but most of the gap remains unexplained. 7Three years of tuition cost, on average, $55,000 for public schools and $107,000 for private schools. In the 2008-2009 academic year, law school students borrowed, on average, $66,000 and $100,000 to attend public and private law schools, respectively. The average values correspond to the American Bar Association, Section of Legal Education and Admissions to the Bar, “Law School Tuition 1985-2008: Average Amount Borrowed For Law School, 2001-2008.”

5

understanding the previously unexplained parts of the earnings gap. In addition, to

understanding the gaps in labor-market outcomes, information on lawyers’ affective

response allows us to question differences in satisfaction on a number of dimensions

and to understand if performance differences are truly choice-driven. While choice may

be constrained by factors such as social norms or inefficiency in the market for

childcare, our analysis of satisfaction offers helpful insight into issues that have been

largely neglected. We see that while there are substantial gaps in performance and, in

turn, earnings, it does not seem that there are gender gaps in satisfaction on important

dimensions. For example, while we find gender differences in areas of specialization,

male and female lawyers are equally satisfied with their choice of specialty. Moreover,

we do not find gender differences regarding satisfaction in work recognition and

opportunities for advancement.

2. Performance Measures in the Legal Profession

The legal profession provides an ideal framework for studying gender

differences in performance. Unlike other high-skilled professions, it uses widely-

accepted objective method to measure and reward lawyers’ productivity. Although the

use of performance pay has increased since the 1970s throughout different economic

sectors, it has become pervasive in professional activities and high-skilled occupations

(Lemieux et al., 2009). In contrast with the legal sector, the methods to measure

performance in other professions and industries are still heterogeneous across firms,

making it difficult to make comparisons within an industry.

2.1. Hours billed

It is standard for law firms in the United States to determine the value of legal

services by computing hourly fees times the number of hours devoted to a case, rather

than by using fixed fees or fees contingent on the outcome of the case (Kritzer, 2004;

Garoupa and Gomez, 2007). Commonly known as billable hours, this method was first

introduced in the 1950s and has become a widely-used management tool within law

firms in the last several decades (American Bar Association, 2002).8

8 The practice of time recording started to become routine in the 1950s (American Bar Association, 2002). By the end of the 1960s, “most mid-sized and large law firms had shifted to hourly billing.”

6

Since billable hours directly determine firms’ revenues, they are also their

preferred way to measure lawyers’ productivity.9 A majority of law firms use billable

hours to determine bonus compensation and have billable-hours requirements for their

associate lawyers (Fortney, 2005; Altman Weil, 2010). To compute the number of hours

a client should pay for, lawyers keep detailed records of the time they devote to each

case (e.g., using time-tracking software). However, it is important to note that the

number of hours that a lawyer bills does not generally coincide with the number of

hours he or she worked. In general, the number of hours lawyers work is larger than the

number of hours billed since there are broad tasks, such as meetings and training

activities, which cannot be assigned to specific clients or cases. Moreover, differences

in work intensity and skills across lawyers may affect how much time each lawyer

needs to have spent in order to bill a given number of hours. For instance, to bill one

hour, two lawyers might have worked different amounts of time, such that the lawyer

taking less time will then be able to take on more assignments and have a higher annual

performance. We can, therefore, interpret hours of work as one of the inputs in the

production function of hours billed.

In this paper, we use the annual number of hours billed by lawyers as the first

measure of lawyers’ performance. As is common in other high-skilled professions (e.g.,

academia, management, etc.), employers are more concerned about overall performance

than about the number of hours worked. Ultimately, it is the annual number of hours

billed that is relevant for law firms since that will determine yearly revenues. In 2006,

the median hourly billing rate for associate lawyers was $200 per hour, and the median

number of hours billed was 1,704 (Altman Weil, 2007). In turn, the median associate

lawyer generated revenue of more than $300,000. There is a great deal of variability

across firms in the billing rates and lawyers’ billable-hours requirements. Typically,

these are increasing in the size of the law firm. There is also variability depending on

the area of law.10

The use of billable hours has proven persistent over the years. Advocates of

billable hours argue that this method serves to calculate the value of the service, to

9A more accurate term is perhaps “perceived productivity” or “perceived performance.” Throughout the paper, we refer to them as “performance” since these are widely-accepted performance measures within the profession, and law firms use them to evaluate lawyers’ annual productivity. 10 The areas of law with larger billing rates are Antitrust, Municipal Finance, Securities, Mergers and Acquisitions, and Intellectual Property. The average number of hours billed also varies across areas: Lawyers working on Trusts and Real Estate, for example, billed 1,507 hours, on average, in 2006 (Altman and Weil, 2007).

7

minimize transaction costs between clients and law firms, and to eliminate uncertainty

and arbitrariness about bonuses for lawyers (American Bar Association, 2002).11 While

the hours billed are accountable, such that they reflect quality and not only quantity,

some critics argue that this method may not reflect all aspects of the services provided

to the client and that it discourages the use of technology that might increase

productivity. Others remark that measuring performance based on hours billed may

induce associate lawyers to overbill clients. However, the Rules of Professional

Conduct of the legal profession establish an ethical code to prevent billing abuse.12 In

addition to ethical incentives, competition between law firms and clients’ participation

constraints reduce the chances that lawyers will overbill.13

2.2. New-client revenue brought to the firm

As a second measure of lawyers’ performance, we consider whether lawyers

personally brought new clients to their firms and how much revenue these new clients

generated. This additional measure captures the quality dimension of lawyers’

performance: Lawyers who provide higher-quality work will establish a good

reputation with clients, who will then be more likely to recommend their services.

Together with hours billed, origination of client revenue—also known as

“rainmaking”—is the most-used objective criterion to measure lawyers’ productivity.

Altman Weil (2010) finds that more than half of law firms –and particularly common in

large ones– use “formal origination credit scoring systems” to reward lawyers’ ability to

attract new clients. Important sources of new client revenue are referrals from previous

clients and from other lawyers (Spurr, 1988; Garicano and Santos, 2004). Therefore,

this performance measure captures ability in creating personal connections, reputation

and visibility. These skills are crucial in promotion decisions because they are

informative about lawyers’ potential performance as law-firm partners.

3. Data Description

11 For a summary of the debate, see American Bar Association (2002). The report argues that “the hourly billing method has endured virulent criticism over the past two decades, [although the criticisms] have not displaced hourly billing or even reduced its dominance as the most common form of law firm billing.” 12 Rule 1.5 of the Model Rules of Professional Conduct, American Bar Association. 13 For a formal model, see, for instance, Garoupa and Gomez-Pomar (2007).

8

Our analysis is done using data from After the JD, a nationally representative

longitudinal survey on lawyers in the United States.14 The After the JD study is a project

of the American Bar Foundation and other legal associations. Lawyers in the sample are

representative of all lawyers who were admitted to the bar for the first time in 2000. The

survey was first conducted in 2002, and the same lawyers were interviewed again in

2007.15 Participants of the survey respond to detailed questions on job characteristics,

employment history, educational background and family status. In 2007 the survey also

included questions on billable hours which is why we focus on this period. They work

primarily in private law practice (54%), government jobs and nonprofit organizations

(25%),16 private industries different from law firms (18%),17 and academic institutions

(3%).18

We focus on lawyers who bill hours—the bulk of those who work in private law

firms. Table 1 reports descriptive statistics for this core sample in 2007. The raw gender

difference in the annual number of hours billed is 152 hours, and the difference in

annual new-client revenue originated is almost $30,000. Assuming that lawyers work 50

weeks per year, they bill 35 hours per week, on average, while working approximately

52 hours per week. Also, on average, male lawyers work five hours more than female

lawyers.19 In total earnings, which refer to lawyers’ reported annual salaries including

bonus components, there is a gender difference of around $17,000. As is commonly

known in the legal profession, total earnings and performance expectations are strongly

positively correlated with the size of the law firm. However, the fraction of female

lawyers working in large organizations is not significantly different from the fraction of

males. There are also no significant gender differences in the average number of years

at the current job. Female lawyers are, however, significantly younger, less likely to be

married and have considerably fewer children. They are also less likely to belong to a

14 Recently, other papers have used After the JD. Oyer and Schaefer (2010) study the salary premium for attending a prestigious law school. Dinovitzer et al. (2009) use the first wave of the data to conduct a detailed descriptive analysis of the gender gap in lawyers' income. 15The response rate in 2002 was approximately 70 percent. Among those responding in 2002, more than 85 percent responded also in 2007. 16This category includes positions such as prosecutor, judge and public defender. 17 This category includes all lawyers working for consulting firms, Fortune 1000 industries, and investment banking. 18Percentages are computed excluding those unemployed and those who did not report their salaries. 19 Hours of work is provided per week rather than per year and refers to the week prior to participating in the survey.

9

minority group.20 The descriptive statistics are very similar when compared to the

lawyers in the sample who do not bill hours.21

The dataset contains detailed educational variables. We use the bracketed

ranking of the institutions that respondents attended as undergraduates and as law

students, as well as their reported grade point average in both institutions.22 We also use

information about whether, as law students, they participated in simulated mock trials

(moot court) and law journals’ editorial activities (general journal and other journal)

since these activities help build skills relevant to practicing law practice and obtaining

jobs. In addition, we also have information on whether respondents held positions as

judicial clerks in state or federal courts. Since judicial clerkships are prestigious

internships through which outstanding students assist judges—usually for the two years

immediately following graduation—having held a position as a clerk captures additional

information about skills. We actually observe that, on average, female lawyers are

significantly more likely than male lawyers to have held judicial clerkships. All of these

education-related variables serve as proxies for ability.

Finally, we also have information on the region in which lawyers lived in 2002.

To account for regional mobility between 2002 and 2007, we construct a variable that

updates the 2002 information if lawyers were last admitted to practice law by a State

Bar’s authority in a different location. Most states require that lawyers pass their

specific bar examination to be able to practice the law in the state. After taking into

account regional mobility, there are thirty regions in the sample, most of which are at

the state level, but for those living in major urban areas, information is disaggregated at

the city level.

4. Gender Gaps in Performance

In this section, we show that there is a large gender gap in performance. We then

investigate alternative explanations for why female lawyers may be neither billing as

much nor raising as much new-client revenue as male lawyers.

From Column 1 of Table 2, we see that male lawyers bill 153 more hours than

female lawyers do, which is equivalent to around ten-percent more hours billed. In

20The study included an oversample of new lawyers from minority groups (black, Hispanic and Asian). 21 The raw earning gap is higher ($25,000) for the overall sample, which seems to be driven by a larger gender gap among those working in professional service firms other than law firms (e.g., investment banking, consulting, etc.) 22 The rankings are based on 1996 and 2003 U.S. News reports for undergraduate and law school studies, respectively.

10

Column 2, we control for individual and firm characteristics, including marriage, age,

the number of children, the presence of children under four years of age, ethnicity, years

of tenure, working full-time, the size of the firm and the type of organization. While

these factors play an important role, they cannot explain the gender differential in hours

billed. In addition, Column 3 shows that including detailed educational variables as

proxies for ability has a negligible effect on the gender gap. Having participated in

journals’ editorial activities, for example, has a positive effect on hours billed, but

including them as controls does not affect the gender gap. Overall, a gap of nearly 100

annual hours remains.

A possible explanation for this gender difference could be unobserved firm

effects that relate to the required number of hours to bill; for example, it could be that

male and female lawyers select into firms that have different requirements. We explore

this using the hours that firms expect their lawyers to bill, which could be related to

gender differences in hiring outcomes or in job assignments. However, from Columns 4

to 6 of Table 2, we find no gender gap in the hours that employers expect female and

male lawyers to bill (with and without controls, respectively).

Regarding the second measure of performance, new-client revenue, we see from

Columns 7, 8, and 9 that male lawyers bring in substantially more revenue than female

lawyers. After controlling for firm and individual characteristics, together with proxies

for ability, the gender gap in revenue is still more than $30,000. Having held a judicial

clerkship has a considerable effect on raising new-client revenue; however, it does not

help explain the gap.

Differences in performance could be due to differences in the amount of work

produced per hour. As mentioned, law firms focus on annual performance in terms of

hours billed to clients. However, studying the ratio of hours billed to hours of work may

help determine whether there are gender differences in productivity per hour worked

and whether female lawyers, perhaps more concerned with the quality than the quantity

of their work, devote more time per billable hour. In Table 3, we find that the gender

coefficient is not significant, implying that female lawyers do not work more hours per

hour billed than males do.

Therefore, the question still remains to be answered: What is causing the gender

differences in performance? To understand the determinants of the gap in performance,

we explore a number of three-principle factors. We start by investigating the traditional

explanations for gender gaps in earnings: discrimination and child rearing. We then

11

investigate whether differences in genders' preferences help to explain part of the gap,

including differences in career aspirations and in area of legal specialization.

4.1. Discrimination

If the employer (partner of the firm) can “interfere” with the number of hours

that the associate lawyer bills, there could be scope for discrimination. In particular, one

could argue that there is some form of discrimination in the assignment of cases since

more-senior colleagues or firm partners typically assign the cases for which associates

bill hours. In order to investigate this possibility, we first study whether or not receiving

enough assignments from the partner explains lower performance. In addition, we study

whether partners might interfere with the way hours billed are measured by discounting

hours or by not giving them full credit.

Table 4 shows the summary statistics for the different reasons that lawyers

report for finding it difficult to meet billable hours. From the table, we do not observe

significant gender differences in the two main responses that could be connected with

discrimination: first, not receiving enough assignments; and second, partners

discounting hours.23 While both explanations seem to be quantitatively important—

accounting for around 30 percent of the difficulty in meeting billable hours—they do

not, on average, affect female lawyers more than male lawyers. In Panel A of Table 5,

we see that not receiving enough assignments does imply that the lawyer bills fewer

hours, suggesting constraints to performance. However, the gender gap remains the

same after including this variable, while the interaction term shows that there is no

gender difference in the hours billed for these “constrained” female and male lawyers.

In other words, a female lawyer who claims not to receive enough case assignments

does not bill less than a male lawyer who claims the same. In Panel B, the results are

similar for partner-discounted hours. Not only does this variable have no effect on the

gender gap, but it also has no significant effect on lawyers’ hours billed in general. In

Columns 4 to 6 of Panels A and B, we look at the gap between hours expected to bill

and actual hours billed. Both reasons have a significant effect on lawyers’ hours billed

beyond what is expected from them; yet, once again, there is no difference between

male and female lawyers and no effect on the gender coefficient.

23 In the following section, we will revisit this table to address some of the other responses. We defer the discussion of these responses until then.

12

One might argue that male and female lawyers have different thresholds for

which they are constrained—i.e., they feel that they do not get enough assignments. If

that is the case, then there may still be scope for discrimination in case assignment. In

Table 6, we see that lawyers billing between 1600 and 1800, between 1800 and 2100 or

more than 2100 hours report being less constrained than those billing 1600 hours or

less.24 The coefficient is significant for only the two upper intervals. In Column 2, when

we interact gender with the different thresholds, we do not find any significant gender

difference. This is reassuring, as it is suggests that the likelihood of being constrained is

the same at the different points in the hours-billed distribution.

Finally, although only descriptive, when lawyers are asked about satisfaction in

the workplace, male and female lawyers are equally satisfied with their advancement in

the firm and with the recognition that they receive for their work. We see that on a scale

of 1 (lowest) to 7 (highest), 61 percent of female and 66 percent of male lawyers rate

their satisfaction with advancement as 5 or more, while 20 percent of both males and

females rate their satisfaction as 3 or below at. Moreover, 68 percent of female and 70

percent of male lawyers rate their satisfaction with recognition for their work as 5 or

higher, while 17 percent of female and 16 percent of male lawyers rate their satisfaction

as 3 or below. Gender differences in levels of satisfaction could suggest discrimination

and, yet, satisfaction levels in both genders appear to be quite equal.

Satisfaction responses also allow us to address whether employers might be

interfering with performance by assigning more routine or less intellectually stimulating

cases to female lawyers. We find that female lawyers are not less satisfied than male

lawyers with the intellectual challenge of their current position ―in fact, male lawyers

are slightly (but significantly) less satisfied than female lawyers in this dimension―. To

explore this issue more in detail we look at lawyers' responses on the frequency of

different types of activities. We do not find gender differences in the amount of time

spent on routine work, on writing motions and depositions, on face-to-face meetings

with clients, in keeping clients updated, and in handling an entire matter on their own.

The only gender difference that we find is that female lawyers report to be more

frequently involved in formulating strategy with more senior attorneys or clients.

Therefore, overall we do not find evidence of female lawyers being held back from

activities that are more challenging or have higher levels of responsibility.

24 These cut-offs are in line with the quartiles of the distribution.

13

4.2. Child Rearing

Gender differences in earnings are often attributed to women having children

and the gender difference in caring for children. We now investigate whether the

presence of children affects performance and whether there is a differential impact on

female lawyers.

Columns 1 and 7 in Table 7 show the gender gaps in hours billed and client

revenue, respectively, controlling for region fixed effects and individual and firm

characteristics, but without controlling for children. In Columns 2, 3, 8 and 9, we see

that neither children nor the presence of young children (i.e., children of preschool age

(under four years old)) has any effect on hours billed or new-client revenue generated,

respectively. In Columns 4 and 10, when we interact the number of children with

gender for each performance measure, we see that there is no differential effect of

children on hours billed or client revenue, respectively. However, Column 5 shows that

there is a differential effect of the presence of young children on billable hours. Having

young children results in female lawyers billing fewer hours but does not affect male

lawyers. In particular, we find that female lawyers with young children bill around 160

fewer hours per year, while male lawyers with young children do not experience a

significant decline in the number of hours billed. This suggests that female lawyers may

shoulder a greater part of the household responsibilities than male lawyers with regard

to raising preschool-aged children. Column 11, however, shows that child rearing does

not help explain the gender gap in new-client revenue. We see that there is no effect of

the presence of either children or young children on raising new-client revenue for

either male or female lawyers.25

4.3. Differences in preferences

To complement more-traditional arguments regarding gender gaps, recent

literature has focused on the effect of gender differences in preferences (see Croson and

Gneezy, 2009). In this section, we study whether possible preference differences

between male and female lawyers help to explain the gender gap in performance. Since

it is not clear in the literature whether these preferences are truly innate or social, we

25 The presence of children of one year of age or below also helps explain the gap in hours billed (but not in new-client revenue); however, the effect is less substantial than the cumulative effect of children under age four.

14

abstract from the debate on the origin of these differences. We focus on whether these

differences determine lawyers’ choices in a way that affects performance. First, we

focus on factors that appear to be crucial in explaining the performance gap in (i.e.,

differences in career aspirations), and then we address other potential factors that do not

appear to play a major role (i.e., choice of area of specialization, willingness to overbill,

networking behavior).

4.3.1. Career Aspirations

Gender differences in the career aspirations of young lawyers may contribute to

the differences in performance. When asked to rate, on a scale from 1 to 10, their

aspirations to become an equity partner within their firm, 60 percent of male lawyers

answered with 8 or more, compared with only 32 percent of female lawyers (see Figure

2). Being able to measure career aspirations is relevant for identification purposes since,

following the career-concerns literature (Fama, 1980; Holmström, 1982, 1999), agents

who put higher weight on their future earnings have stronger incentives to put in effort,

which affects performance. This is particularly true for workers at an early stage of their

careers since the incentive to perform better increases with the level of uncertainty

about workers’ skills. Even in the presence of explicit monetary rewards for

performance, such as bonus compensation, career concerns may play a considerable role

in workers’ effort decisions (Gibbons and Murphy, 1992).

Columns 3 and 7 of Table 8 show that aspirations have a strong positive effect

on the hours billed and the new-client revenue generated.26 Interestingly, while

differences in aspirations do not fully explain the gender differences in hours billed

(Column 3), they do explain differences in new-client revenue since the gender

coefficient is no longer significant when we control for aspirations (Column 7). This

suggests that the gender differences in aspiration levels can explain the remaining

gender difference in new-client revenue generated by lawyers. This is intuitive, as new-

client revenue can be seen as lawyers' long-term investment in their firms. Initiating and

finding new clients requires time and effort, but career concerns may make this

worthwhile. Interestingly, from Columns 4 and 8, we see that there is no differential

effect of aspirations on hours billed and client revenue, respectively. In other words, if

26 Not all lawyers responded to the question on aspirations. In Columns 1, 2, (5, 6), we see that there is little difference in the gender coefficient for the different samples on hours billed (new-client revenue).

15

male and female lawyers aspire at the same level, there is no difference in the hours

they bill or revenue they generate.

One may argue that the aspirations variable captures the likelihood of becoming

partner, as the question is asked of lawyers in their seventh year out of law school. This,

they may already have an idea about their chances of becoming partner, which usually

takes place between the seventh and tenth year after passing the bar examination. We

deal with this concern by using the instrumental variable technique. We instrument for

aspirations using two different variables from the first wave of the survey, which was

conducted in 2002, when the lawyers were only two years out of law school. The

instruments we use are: 1) How satisfied are you with your decision to become a

lawyer? 2) How much longer would you like to stay with your current employer? The

instruments capture the lawyers’ contentment within the industry and within the firm,

respectively. We expect respondents, after only two years of working as lawyers, to still

have a high degree of uncertainty regarding future performance and the likelihood of

becoming a partner, and yet correlated with aspirations. In the first stage, both

instruments are strong and are individually significant at the one-percent level, with

respect to aspirations. The second stage is given in Columns 9 to 14 of Table 8. In line

with our OLS findings, we see that aspirations do play an important role in explaining

the gender differences in client revenue. While aspirations are important for the number

of hours billed, they do not explain the gender difference in billable hours.

4.3.2. Other Preferences-Related Explanations

In this subsection, we look at other differences in preferences that could

potentially affect performance. Overall, we find that while there may exist important

gender gaps in these factors, they contribute very little to explaining performance gaps.

First, we explore gender differences in billing behavior. From Table 4, we saw

that one reason why lawyers find it more difficult to meet billable hours is that they feel

that they are less likely to bill for actual hours worked, as compared to their colleagues.

In response to this, we see that female lawyers are four-percent more likely than male

lawyers to select this reason. It is possible that female lawyers may be less willing than

their male colleagues to overbill clients. For instance, if women were more risk-averse

than men (Eckel and Grossman, 2008), the possibility of being caught by the client or

by a partner would be a stronger incentive for female lawyers not to overbill. Similarly,

if men are more assertive and less empathic (Ruble et al., 2006), they might have a

16

higher tendency to overbill. While overbilling is likely to exist, the Rules of

Professional Conduct of the legal profession establish an ethical code to prevent such

abuse. Moreover, competition between law firms and clients’ participation constraints

reduce the chances for lawyers to overbill (see Section 2 for more discussion on

overbilling). A second reason could be gender differences in how male and female

lawyers value their work. Barron (2003) and Major et al. (1984) find that women may

feel less entitled to higher compensation.

A more thorough analysis of differences in billing behavior shows that, unlike

career aspirations, it does not explain the gender differences in performance. In

particular, from Table 9, we see that lawyers who report that they are less likely than

their colleagues to bill for actual hours worked do bill fewer hours. Nevertheless, the

gender gap persists, and the interaction with gender is insignificant—suggesting that

male and female lawyers who respond in the same way do not differ in the hours they

bill. These results hold for the gap between expected hours and actual hours billed.

In addition, we find that the other possible explanations listed in Table 4 do not

have a significant effect on the gender gap. There is no gender difference in the hours

billed for those who report difficulties in meeting billable hours due to personal choice

or due to spending too much time on pro bono or administrative tasks. Regarding health

issues, we do observe in Table 4 that female lawyers are 12-percent more likely than

males to select this reason. This could be in line with Ichino and Moretti (2010), who

find a connection between women’s absenteeism and the female menstrual cycle. In our

analysis, however, health issues with regard to difficulty in billing more hours do not

appear to have an effect on either the gender gap or performance.

A second potential explanation is related to preferences in networking behavior.

The willingness to spend time attending networking functions and or participating in

recreational activities with other lawyers or clients for networking purposes may differ

by gender. On average, in a typical week, male lawyers work 15 percentage points more

than female lawyers; they attend networking events 11 percentage points more and are

40 percentage points more likely to participate in recreational activities (e.g., golf) for

networking purposes with other lawyers or clients. Nevertheless, from Table 10, we do

not find that these differences are a relevant source of the gender gap in performance.

Networking could affect the gender gap in performance in two ways: first, if female

lawyers devote less time to networking; and second, if networking affects male and

female lawyers differently. For instance, the previous literature found differences in the

17

type of networks that male and female managers build (Ibarra, 1997). From Table 10,

we see that networking does not affect hours billed but has important consequences for

raising new-client revenue. An extra hour spent networking is associated with raising an

additional $2,800. However, Column 5 shows that controlling for networking does not

reduce the gender coefficient for new-client revenue. Thus, the amount of time devoted

to networking does not explain the performance gap. In addition, we analyze whether

networking affects male and female lawyers differently for a given number of

networking hours. In Columns 3 and 6, the interaction term of networking with gender

is not significant for either hours billed or client revenue. Therefore, an extra hour spent

networking has the same performance return for male and female lawyers.

We obtain similar results for working on weekends. In Table 11, Columns 2 and

4 show that time spent working on weekends has important consequences for both hours

billed and client revenue. In particular, one additional weekend hour worked per week is

associated with an increase of 14 hours billed per year and an additional $2,800 in new-

client revenue. Although time worked on weekends has a substantial effect on

performance, it does not seem to explain the gender gap in performance. Moreover, time

worked on weekends does not affect female and male lawyers differently, as shown in

Columns 3 and 6.

Finally, we explore whether a lawyer's specialty affects performance. Although

our results are within the same profession, one may relate differences in the area of

specialization to the occupational segregation literature.27 We control for lawyers’ area

of specialization by using the percentage of their time that respondents devote to 27

different areas of law listed in the survey. Although we do find gender differences in the

areas of specialization, they do not seem to be relevant for the differences in

performance. Out of the 27 specialties listed, we find that, compared to the overall

sample, female lawyers are more significantly represented in Family Law, Probate

(Wills and Trusts), Employment Law (Management), and Public utilities and

Administrative Law, while Intellectual Property and Criminal Law have a significantly

larger number of male lawyers.28 Nevertheless, Table 12 shows that controlling for the

27 While occupational segregation has declined over the years, there still appears to be a tendency for women and men to choose different types of jobs and different specialized training within the same profession See, for instance, Blau et al. (1988), Goldin (1990), Blau and Kahn (2000) and Bertrand et al. (2010). 28 Lawyers in the sample report the percentage of time that they devote to each of the legal areas. The results provided are robust to different possible ways in which to aggregate this information. We do not find either men or women to be overrepresented in the remaining areas of specialization: General

18

area of specialization does not have a substantial effect on the gender gap on

performance. We see that the gender coefficient decreases slightly for hours billed

(Column 2), while it increases slightly for client revenue (Column 4).

5. The Role of Performance on the Earnings Gender Gap

Given that there exist considerable differences in performance, in what follows,

we analyze how these differences translate into differences in earnings. We find that

while traditional individual and firm controls explain around 50 percent of gender

earning differences, performance measures explain almost the entire remaining gap. We

present the results comparing the analysis with and without controlling for performance

measures. Before proceeding, the next subsection exposes the identification problems

that may arise when performance is not observed.

5.1. Performance and the Omitted-Variable Problem

In a standard competitive model, lawyers’ compensation would be determined

by their performance (e.g., W=P). Therefore, in the following econometric specification

for annual earnings,

0 1 2 ,i i i iW Gender P

the coefficient for Gender would be different from zero only if, given the same

performance, P, male and female lawyers are still paid differently. Therefore, β1 could

be interpreted as discrimination in earnings. In other words, if earnings differences

persist after controlling for performance—together with individual, firm and location

characteristics—then there could be evidence that lawyers who perform equally are

compensated differently.

A persistent problem in the gender-gap literature is that productivity or

performance is typically not observable, especially when studying high-skilled sectors

and professions, thus implying an omitted-variable problem. Moreover, whenever

performance measures are not available, a common strategy in the literature is to use

observable characteristics to proxy for performance:

Practice, Antitrust, Bankruptcy, Civil Litigation, Civil Rights, Commercial Law, Employment Law (Unions), Environmental Law, General Corporate Law, Immigration Law, Municipal Law, Personal injury (Plaintiff), Personal Injury (Defense), Real Estate (Commercial), Real Estate (Personal), Securities, Tax Law and ‘Other’ areas.

19

0 1 2 ,i i i iP Gender X u

where the proxies for performance would be Gender and Xi (e.g., child-rearing-related

variables, firm characteristics, educational background, etc.).

Introducing the performance equation into the expression for annual earnings,

we see that the coefficient for Gender might be significant due to discrimination or to

gender differences in performance not related to Xi. In particular, rewriting the

expression as:

0 2 0 1 2 1 2 2 2( ) ( ) ( ),i i i i iW Gender X u

we cannot distinguish between β1 and β2 γ1. In other words, we cannot differentiate

part of the coefficient on Gender due to discrimination from that due to differences in

performance.

As discussed in Section 2, the legal profession is distinctive in that there do exist

recognized ways in which performance is measured—namely, Hours billed and New-

client revenue. In turn, we are able to control directly for performance—hence,

alleviating the omitted-variable bias problem.

5.2. Gender Gap in Earnings (without Performance Measures)

We start by estimating (log) annual earning equations using ordinary least

squares. The estimations shown in Table 13 are done with and without controlling for

individual and firm characteristics.

The raw gap in mean log earnings between male and female lawyers for the full

sample is 25 log points (Column 1). In Column 2, we control for individual

characteristics, including marriage, age, the number of children, the presence of children

under age four, ethnicity, years of tenure, and working full-time. The inclusion of these

characteristics explains a substantial fraction of the gender gap; however, 18 log points

are still unexplained. Marriage and the presence of children do not seem to directly

affect log earnings, but working full-time instead of part-time and the years of tenure do

affect wages. Note that if we use weekly hours worked instead of full-time status, we

find a similar effect on the gender gap (Column 3). Age appears to have an effect on log

earnings; however, since all workers are of the same cohort, there is little variation in

age. When we add the quadratic terms, it is no longer significant.

20

In Column 4, we control for important firm characteristics: the size of the firm

and the type of organization. While these factors play an important role—the gap falls

to 12.6 log points—they cannot fully explain the gender earning differential. In general,

working in a larger firm, working in a private law firm, or working in the private sector

in general all correspond to higher earnings.

The individual and firm characteristics together explain 50 percent of the raw

gender gap, but the other 50 percent remains unexplained. Interestingly, Wood et al.

(1993), in a study of University of Michigan Law School graduates from the class of

1972-75, also find a gender gap in annual earnings of 12.4 log points when controlling

for similar characteristics. The proportion of female lawyers in the 1970s was

considerably lower; in their study, female lawyers comprise only nine percent of the

sample.

5.3. Gender Gap in Earnings (with Performance Measures)

In this section, we analyze the effect of performance on earnings differences by

focusing on lawyers who bill hours. Among those who bill hours, more than 93 percent

work in law firms, and the remaining lawyers work in solo practices. The gender gap in

earnings for those who bill hours is quite similar to the gap in other sectors. In Table 14,

we show that there is an overall gap of 12.6 log points, a slightly lower gap of 10.6 log

points if we consider those who work in the public sector, and 10.0 log points for those

who bill hours. There may be selection of more-able women into private law firms,

which reduces the raw gender gap; however, the gap does not disappear. We will

address the selection issue later in this section.

In Table 15, we include the main performance variables: hours billed and the

amount of new-client revenue generated. To compare the results, in Column 1, we show

the gender gap controlling only for individual and firm characteristics. Column 2 shows

that when we include region fixed effects, the gap is largely unchanged. Controlling for

performance (columns 3, 4, and 5) explains a considerable part of the remaining gender

gap. In particular, the number of hours billed has a strong and positive effect on

earnings; we find that billing 100 additional hours per year leads to a 3.1-percent

increase in salary. Interestingly, the inclusion of this variable reduces the gender earning

gap to 6.5 log points. Furthermore, the importance of hours billed in calculating

lawyers' earnings is clearly noticeable when comparing the earnings regression without

controls in Column 6 with the same regression after controlling for hours billed in

21

Column 7. The single inclusion of hours billed makes the R-squared increase from two

percent to 26 percent.

Column 4 shows that client revenue, too, has an important effect on annual

earnings. Bringing in new-client revenue worth $100,000 implies an increase of around

4.5 percent in earnings. From Column 5, we see that when including both performance

measures, the gender gap in earnings falls to 5.5 log points and is significant only at the

ten-percent level. In Columns 10 and 11, we include the squared and cubic terms,

respectively. There seems to be a nonlinear relationship of these variables, but it does

not affect the gender coefficient. Overall, the analysis shows that it is key to control for

differences in workers’ performance.

To study the difference in earnings per unit of performance, we show that there

is no gender difference in the reward for each hour billed or for each dollar of client

revenue raised by the lawyers. Since earnings per hours billed serves a proxy for the

hourly rates charged to clients, this regression serves as an additional way to study

explicit discrimination. Columns 1 and 2 in Table 16 show that the gender coefficient is

not significant for earnings per hour billed and earnings divided by client revenue

generated, respectively.

5.4. Robustness Checks

Hours billed and new-client revenue are good summary statistics for true

productivity. In this section, we show that controlling for other things that may be

correlated with performance does not contribute much to our main findings.

First, we control for a wide range of education variables that proxy for ability. In

Column 3 of Table 17, we control for undergraduate university ranking, law school

ranking, having held a judicial clerkship or having participated in Law Review

(member/editor) or Moot Court (member/leader) activities while in law school. While

some of the variables are significant after controlling for other individual and firm

characteristics, they neither change the gender coefficient nor help to explain the gender

gap. The positive and significant effect of law school ranking is consistent with Oyer

and Schaefer's (2010) findings that attending a prestigious school has a considerable

effect on annual salary.

Second, we investigate the effect of area of specialization on earnings. In

Column 4 of Table 17, we show that while areas of specialization have some effect on

gender gap, when comparing Columns 2 and 4, we see that they do not affect the

22

coefficients or the significance of the performance measures—hours billed and client

revenue.

Finally, we address the possibility of selection differences between men and

women into billing hours. In Table 18, we find that female lawyers are, on average,

three-percent less likely to enter a job that requires billing hours (Column 1). However,

we find that it is the more-able lawyers, rather than less-able lawyers, who tend to select

into jobs that bill hours, as shown in Column 2. From Column 3, we see that this is

equally true for male and female lawyers, such that we can rule out that more-able

women are self-selecting themselves out of jobs that require billing hours. In other

words, lower hours billed by female lawyers do not seem to be due to a selection of

less-able women into jobs requiring billing hours. In fact, we do observe that the gap is

slightly smaller when focusing only on lawyers who bill hours.

6. Conclusion

We have examined gender differences in performance among high-skilled

individuals. Using the legal framework, in which there are well-defined and

homogeneous performance measures, we find a substantial gender gap in performance.

These gaps appear to be consequential since we find that the difference in earnings

among male and female young lawyers is strongly related to gender gaps in

performance.

We explore three principal factors that can explain gender gaps in performance:

(i) discrimination in the workplace; (ii) the presence of children in the household—

particularly young children; and (iii) preference-related factors. An important finding is

that discrimination in law firms—whereby senior lawyers (i.e., law-firm partners) have

some scope to interfere with variables associated with performance—does not seem to

explain the gaps. While the presence of pre-school children contributes, in part, to the

gaps in performance, it is not the only key determinant. Aspirations to become a partner

and, perhaps, more-general career concerns explain an important part of the gender gap.

Gender differences exist on other dimensions, such as areas of specialization, time spent

networking, and time spent working on weekends. While these factors influence

performance, they do not appear to explain the gender gaps in performance.

The differences in performance have important consequences. We show that

after controlling for detailed individual- and firm-level characteristics, 50 percent of the

23

gender gap in earnings remains unexplained. Traditionally, the lack of data on key

variables such as performance, especially in skilled or non-manual jobs, would entail

speculating on what can explain the remaining gap. We show that the inclusion of an

important omitted variable can explain a large part of the remaining gap. A relevant

implication of these results is that gender earnings inequality might increase in the near

future due to the growing number of high-skilled workers explicitly compensated based

on performance.

An important next step is to examine more deeply why career aspirations and the

effects of raising children differ between high-skilled females and their male

counterparts. While we show that discrimination at the firm level does not seem to be an

important determinant, it may be that social norms or some other type of social

pressures burden even the most elite professional women. We find similar levels of

satisfaction in the workplace among female and male lawyers, which may suggest that

these differences are truly choice-driven. Finally, further research might address

whether the use of objective performance measures permits the elimination of possible

discrimination channels.

References

Altman Weil (2007), “Survey of Law Firm Economics,” Altman Weil, Inc. Altman Weil (2010), “Compensation Plans for Law Firms,” edited by J.D. Cotterman, American Bar Association Books. Altonji, J. G. and R. M. Blank (1999), "Race and Gender in the Labor Market", in Orley Ashenfelter and David Card (eds), Handbook of Labor Economics, Volume 3C. Handbooks in Economics, vol. 5, Amsterdam: North-Holland, pages 3143-3259 American Bar Association (2002), “Commission on Billable Hours Report.” American Bar Association (2006) “Charting Our Progress: The Status of Women In the Profession Today.”

Barron, L., (2003) “Ask and you shall receive? Gender differences in negotiators’ beliefs about requests for a higher salary,” Human Relations, vol. 56, pp. 635–662.

Bertrand, M, Goldin, C. and Katz, L. F (2010), "Dynamics of the Gender Gap for Young Professionals in the Financial and Corporate Sectors," American Economic Journal: Applied Economics, vol. 2, pp. 228-55. Blau, F.D. and Kahn, L. M. (2000), “Gender Differences in Pay,” Journal of Economic Perspectives, vol. 14, pp. 75-99.

24

Blau, F. D., Simpson, P., and Anderson, D. (1988), “Continuing progress? Trends in the occupational segregation in the United States over the 1970s and the 1980s,” Feminist Economics 4(3), 29–71. Croson,R. and Gneezy, U. (2009), “Gender Differences in Preferences”, Journal of Economic Literature, vol. 47, pp. 1-27. De la Rica, S., Dolado, J.J., and Vegas, R. (2010), “Performance Pay and the Gender Wage Gap: Evidence from Spain,” working paper. Dinovitzer, R., Reichman, N., and J. Sterling (2009). "The Differential Valuation of Women's Work: A New Look at the Gender Gap in Lawyers' Incomes," Social Forces, 88:819-854.

Eckel, C. C. and P. J. Grossman, P. J. (2008) “Sex and risk: Experimental evidence,” Handbook of Experimental Economics Results, ch. 113, pp. 1061-1073.

Fama, E.F. (1980), "Agency Problems and the Theory of the Firm," Journal of Political Economy, vol. 88, pages 288-307. Fortney, S. S. (2005), “In Pursuit of Attorney Work-Life Balance: Best Practices in Managament,” NALP Foundation. Fox, L.J. (2007), “Raise the Bar: Real World Solutions for a Troubled Profession,” American Bar Association. Garicano, L. and Santos, T. (2004), “Referrals,” American Economic Review, vol. 94, pp. 499-525 Garoupa, N. and Gomez-Pomar, F. (2008), “Cashing by the Hour: Why Large Law Firms Prefer Hourly Fees over Contingent Fees,” Journal of Law, Economics and Organization, vol. 24, pp. 458-475. Gibbons, R. and Murphy, K.J. (1992), "Optimal Incentive Contracts in the Presence of Career Concerns: Theory and Evidence," Journal of Political Economy, vol. 100, pp. 468-505. Goldin, C. (1990), “Understanding the Gender Gap: An Economic History of American Women.” New York: Oxford University Press. Heywood, J. S. and Parent, D. (2009) “Performance Pay And The White-Black Wage Gap,” Departmental Working Papers 2009-07, McGill University, Department of Economics. Holmström, B. (1982, 1999), "Managerial Incentive Problems: a Dynamic Perspective," Review of Economic Studies, vol. 66, pp. 169-183.

Ibarra, H. (1997), “Paving an alternative route: Gender differences in managerial networks,” Social Psychology Quarterly, vol. 60, pp. 91–102.

25

Ichino, A. and Moretti, E. (2010) “Biological Gender Differences, Absenteeism and the Earnings Gap,” American Economic Journal: Applied Economics, vol. 1, 183-218. Kritzer, H.M. (2004), “Risks, Reputations, and Rewards: Contingency Fee Legal Practice in the UnitedStates. Palo Alto, Calif.: Stanford Law and Politics Partnerships in the Am Law 200,” North Carolina Law Review, 84, pp. 1691-1750 Lemieux, T., MacLeod, W. B. and Parent, D. (2009), “Performance Pay and Wage Inequality,” Quarterly Journal of Economics, 124, pp. 1-49.

Major, B., D. McFarlin, and D. Gagnon (1984) “Overworked and underpaid: On the nature of gender differences in personal entitlement,” Journal of Personality and Social Psychology, vol. 47, pp. 1399–1412.

Manning, A., and J.Swaffield (2008), “The Gender Gap in Early-Career Wage Growth.” Economic Journal, 118(530): 983–1024. Oyer, P. and Schaefer, S. (2010), “The Returns to Attending a Prestigious Law School,” working paper. Rosen, S. (1992) "The Market for Lawyers," Journal of Law and Economics, vol. 35, No.2., pages 215-246.

Ruble, D. N., Martin, C. L., and Berenbaum, S.A. (2006) “Gender development,” Handbook of child psychology, ch. 5, pp. 858-932, New York: John Wiley & Sons.

Spurr, S.J. (1988) “Referral Practices Among Lawyers: A Theoretical and Empirical Analysis,” Law & Social Inquiry, 12, 87-109. Wood, Corcoran, and Courant (1993) Pay Differences among the Highly Paid: The Male-Female Earnings Gap in Lawyers' Salaries, Journal of Labor Economics, vol. 11, No. 3.

26

Tables and Figures

TABLE 1 – DESCRIPTIVE STATISTICS

Male Lawyers Female Lawyers

Obs. Mean Std. Dev. Obs. Mean Std. Dev. P-

Value Total Earnings ($) 684 150,667 74,531 441 132,685 70,282 0.00 Hours billed (annual) 684 1,826 535 441 1,677 520 0.00 New Client Rev. ($) 684 53,346 171,965 441 23,349 68,892 0.00 Age (years) 684 36.12 4.98 441 35.29 4.92 0.01 Marriage 684 0.81 0.39 441 0.75 0.43 0.02 Children 684 1.22 1.24 441 0.82 0.91 0.00 White 684 0.83 0.38 441 0.75 0.43 0.00 Hours worked (per week) 684 54.09 12.80 441 48.83 13.84 0.00 Tenure (years) 684 5.18 2.49 441 5.26 2.44 0.59 Private Law Firm 684 0.92 0.27 441 0.93 0.26 0.57 Size of workplace > 100 684 0.48 0.50 441 0.51 0.50 0.26 Law School Ranking 597 4.95 1.08 392 5.05 1.10 0.17 UG Uni Ranking 662 12.89 3.50 435 13.04 3.62 0.48 Judicial Clerk 684 0.02 0.15 441 0.03 0.17 0.44 Moot Court 684 0.32 0.47 441 0.35 0.48 0.31 General Journal 684 0.22 0.42 441 0.20 0.40 0.39 Specific Journal 684 0.20 0.40 441 0.25 0.44 0.05

Notes: Total Earnings are calculated as a sum of straight salary and bonus. Hours billed (annual) is the number of hours billed last year (2006). New Client Rev is the approximate amount of new-client revenue (expressed in U.S. dollars) generated last year (2006). Marriage takes the value one if the lawyer is married, remarried after divorce or in a domestic partnership and zero if single, divorced or separated, widowed, or other. White takes the value one if the lawyer is Caucasian and zero if lawyers are from minority groups (black, Hispanic and Asian). Private Law Firm takes the value one if the lawyer works in a private law firm and zero if the lawyer works for another organization (solo practice, federal government, state or local government, legal services or public defender, public interest organization, educational institution, professional service firm, other Fortune 1000 industry/service, other business/industry, labor union, trade association, others). Size of workplace > 100 takes the value one if the number of people employed in the organization is greater than 100 and zero otherwise. Hours worked (per week) is the number of hours spent working last week (at the office or away from the office). Undergraduate Uni Ranking and Law School Ranking are bracketed rankings based on U.S. News reports of 1996 and 2003 for undergraduate and law school studies, respectively. Both variables are redefined such that the higher the value, the more prestigious is the educational institution. Judicial Clerk takes value one if the lawyer has held a position as a judicial clerk in state or federal courts and zero otherwise. Moot Court takes the value one if the lawyer participated in simulated mock trials as a student and zero otherwise. General (Specific) Journal takes value one if the lawyer participated in law journals’ editorial activities as a student and zero otherwise.

27

TABLE 2 – PERFORMANCE GAPS

Hours Billed Expected Hours Bill New Client Rev.

[1] [2] [3] [5] [6] [7] [8] [9] [10]

Female -0.153*** -0.105*** -0.0981*** -0.081 -0.0562 -0.0442 -0.299*** -0.289*** -0.317***

[0.0329] [0.0315] [0.0319] [0.0502] [0.0462] [0.0474] [0.0916] [0.101] [0.104]

Age -0.0129*** -0.0112*** -0.0115** -0.0120** -0.0085 -0.011

[0.00315] [0.00331] [0.00477] [0.00503] [0.0101] [0.0108]

Married 0.0658* 0.0706* 0.037 0.0394 0.251** 0.254*

[0.0391] [0.0396] [0.0581] [0.0592] [0.126] [0.129]

No. Children -0.0266 -0.0275* 0.0269 0.0281 -0.0382 -0.05

[0.0163] [0.0167] [0.0240] [0.0247] [0.0525] [0.0547]

Child Aged <4 -0.000425 0.00333 -0.0419 -0.0571 -0.105 -0.129

[0.0384] [0.0390] [0.0567] [0.0578] [0.123] [0.128]

White -0.0113 -0.0268 0.00408 -0.00405 0.0343 0.0183

[0.0377] [0.0388] [0.0577] [0.0598] [0.121] [0.127]

Tenure 0.0147** 0.0129** -0.00707 -0.00981 0.0388** 0.0407**

[0.00588] [0.00601] [0.00885] [0.00911] [0.0189] [0.0197]

Full-Time 0.490*** 0.492*** 0.321*** 0.314*** 0.161 0.126

[0.0618] [0.0624] [0.0919] [0.0936] [0.199] [0.204]

UG Uni Ranking -0.00154 -0.00541 -0.0114

[0.00422] [0.00651] [0.0138]

Law School Ranking 0.00977 -0.0302 0.0442

[0.0156] [0.0232] [0.0511]

Judicial Clerk 0.116 0.106 0.724**

[0.0886] [0.126] [0.290]

Moot Court 0.0103 0.0772* 0.0699

[0.0300] [0.0460] [0.0983]

General Journal 0.0846** 0.0673 -0.0103

[0.0351] [0.0516] [0.115]

Specific Journal 0.0765** 0.0354 -0.0085

[0.0351] [0.0525] [0.115]

Constant 1.842*** 0.666 0.588 1.566*** 1.439** 1.709** 0.527*** 0.636 0.342

[0.0205] [0.477] [0.486] [0.0317] [0.696] [0.718] [0.0571] [1.510] [1.591]

Firm Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes

Region Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes

Observations 1,039 1,014 974 803 800 770 1,039 1,014 974

R-squared 0.02 0.30 0.31 0.00 0.32 0.34 0.01 0.07 0.08

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level. Expected Hours Bill is the annual hours (expressed in 1000 hours) the lawyer was expected to bill last year (2006). Hours billed is the annual number of hours billed (expressed in 1000 hours) last year (2006), and New Client Rev is the approximate amount of new-client revenue (expressed in 100,000s of U.S. dollars) generated last year (2006). Firm controls include the type of organization (solo practice, private law firm, federal government, state or local government, legal services or public defender, public interest organization, educational institution, professional service firm, other Fortune 1000 industry/service, other business/industry, labor union, trade association, others) and the size of organization, which are bracketed (0-5, 6-10, 11-25,25-50,51-100,101-150,151-200,201-250,251-500,501-1000,1000+). For definitions of other variables, see Table 1.

28

TABLE 3 – RATIO OF HOURS WORKED TO HOURS BILLED

Hours Billed/Hours

Worked Female 0.013 [0.0685] Age 0.000833 [0.00686] Married -0.00568 [0.0851] No. Children 0.013 [0.0355] Child Aged <4 0.0202 [0.0835] White 0.0406 [0.0820] Tenure -0.00376 [0.0128] Full-Time -0.564*** [0.135] Constant -0.335 [1.038] Firm Controls Yes Region Fixed Effects Yes Observations 1009 R-squared 0.04

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level. We calculate the annual hours of work, assuming a 50-week work year.

29

TABLE 4 – REASONS FOR DIFFICULTIES IN MEETING BILLABLE HOURS

Male Lawyers Female Lawyers

Obs. Mean Std. Dev. Obs. Mean Std. Dev. P-Value

Not enough Assignments 662 0.23 0.42 453 0.26 0.44 0.35

Partner discounted hours (or did not give them full credit) 660 0.13 0.34 451 0.14 0.35 0.98

Personal choice 664 0.36 0.48 451 0.39 0.49 0.05

Health issues 659 0.07 0.26 447 0.19 0.39 0.00

Less likely to bill for actual hours worked compared to colleagues 657 0.19 0.39 444 0.23 0.42 0.12

Too much time spent on pro bono 658 0.05 0.22 446 0.05 0.22 0.97

Too much time spent on administrative tasks 661 0.37 0.48 451 0.38 0.49 0.74

Notes: The seven dummy variables in the table are constructed based on the survey question: “Which of the following posed difficulties in meeting your billables in 2006?” Respondents could choose as many reasons as applicable to their case. By “meeting your billables,” we refer to reaching the annual number of hours billed required in the corresponding firm to obtain an annual bonus.

30

TABLE 5 – PERFORMANCE: DISCRIMINATION

PANEL A Hours Billed (Hours Billed-Exp. Hours Bill) New Client Rev.

[1] [2] [3] [4] [5] [6] [7] [8] [9]

Female -0.105*** -0.102*** -0.0939*** -35.07 -30.94 -44.66 -0.289*** -0.287*** -0.308***

[0.0315] [0.0312] [0.0342] [46.10] [45.72] [51.30] [0.101] [0.101] [0.111]

Not Enough Assignments -0.154*** -0.137*** -187.4*** -211.2*** -0.131 -0.177

[0.0369] [0.0473] [50.05] [64.27] [0.120] [0.154] Female* Not Enough Assig. -0.044 59.15 0.114

[0.0736] [100.1] [0.239]

Constant 0.666 0.682 1.395*** 176.3 162.8 164.1 0.636 0.625 0.629

[0.477] [0.473] [0.465] [694.0] [688.0] [688.3] [1.510] [1.510] [1.511]

Individual Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes

Firm Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes

Region Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes

Observations 1,014 1,014 1,014 800 800 800 1,014 1,014 1,014

R-squared 0.30 0.31 0.31 0.11 0.12 0.12 0.07 0.07 0.07

PANEL B Hours Billed (Hours Billed-Exp. Hours Bill) New Client Rev.

[1] [2] [3] [4] [5] [6] [7] [8] [9]

Female -0.105*** -0.105*** -0.0922*** -35.07 -36.24 -63.93 -0.289*** -0.289*** -0.292***

[0.0315] [0.0315] [0.0331] [46.10] [45.92] [48.88] [0.101] [0.101] [0.107]

Partner Discount Hours -0.0335 0.0185 -171.1*** -257.1*** -0.143 -0.152

[0.0484] [0.0626] [64.52] [83.15] [0.156] [0.202] Female* Partner Dis. Hours -0.126 211.2 0.0235

[0.0962] [129.0] [0.310]

Constant 1.411*** 1.416*** 1.411*** 176.3 250.8 -509.3 0.636 0.182 0.658

[0.469] [0.469] [0.469] [694.0] [691.8] [691.3] [1.510] [1.537] [1.511]

Individual Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes

Firm Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes

Region Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes

Observations 1,014 1,014 1,014 800 800 800 1,014 1,014 1,014

R-squared 0.30 0.30 0.30 0.11 0.11 0.12 0.07 0.07 0.07

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level. Hours Billed and Expected Hours to Bill are expressed in 1000 hours. New Client Revenue is expressed in 100,000 U.S.dollars. Not Enough Assignments takes value one if the lawyer responds that not enough assignments are the reason for why he or she had difficulty meeting billables and zero otherwise. Partner Discounted Hours takes value one if the lawyer responds that partner-discounted hours (or lack of full credit) is the reason why he or she had difficulty meeting billables and zero otherwise.

31

TABLE 6 – PERFORMANCE: DISCRIMINATION

(GENDER DIFFERENCES IN CONSTRAINT THRESHOLD)

Constrained [1] [2] Female -0.0087 -0.0106 [0.0268] [0.0552] 1600<Hours Billed<1800 0.0405 0.0416 [0.0389] [0.0533] 1800<Hours Billed<2100 -0.130*** -0.130*** [0.0375] [0.0489] 2100<Hours Billed<3000 -0.255*** -0.260*** [0.0422] [0.0537] Female*(1600<Hours Billed<1800) -0.003 [0.0751] Female*(1800<Hours Billed<2100) -0.0013 [0.0681] Female*(2100<Hours Billed<3000) 0.0151 [0.0803] Constant -0.171 -0.173 [0.395] [0.399] Individual Controls Yes Yes Firm Controls Yes Yes Region Fixed Effects Yes Yes Observations 1,014 1,014 R-squared 0.12 0.12

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level. The dependent variable, Constrained, takes the value of 1 if the individual responds that she does not have enough assignments (see notes in Table 12). Hours Billed is expressed in 1000 hours. The omitted category of 800<=Hours Billed<=1600, where 800 is the lowest number of hours billed in our sample. Each category represents quartiles in the hours-billed distribution.

32

TABLE 7 – PERFORMANCE: CHILD-REARING

Hours Billed New Client Rev. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12]

Female -0.0959*** -0.105*** -0.105*** -0.0857** -0.0317 -0.0526 -0.265*** -0.286*** -

0.289***-

0.325** -

0.309** -

0.326** [0.0311] [0.0314] [0.0315] [0.0407] [0.0384] [0.0416] [0.1000] [0.101] [0.101] [0.131] [0.124] [0.135] No. Children -0.0267* -0.0266 -0.0219 -0.0288* -0.0392** -0.0608 -0.0382 -0.0467 -0.0376 -0.0461 [0.0140] [0.0163] [0.0174] [0.0162] [0.0180] [0.0452] [0.0525] [0.0562] [0.0526] [0.0585] Children aged < 4 -0.000425 0.00249 0.0712 0.0845* -0.105 -0.11 -0.124 -0.113 [0.0384] [0.0386] [0.0439] [0.0450] [0.123] [0.124] [0.142] [0.146] Female*No. Children -0.0227 0.0472 0.0409 0.0384 [0.0298] [0.0358] [0.0960] [0.116] Female*Childr. aged < 4 -0.203*** -0.258*** 0.054 0.00926 [0.0615] [0.0743] [0.199] [0.241] Constant 0.603 0.665 0.666 0.645 0.682 0.73 0.495 0.146 0.165 0.645 0.647 0.199 [0.476] [0.477] [0.477] [0.478] [0.475] [0.476] [1.500] [1.536] [1.536] [1.511] [1.511] [1.542] Individual Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Firm Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Region Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 1,014 1,014 1,014 1,014 1,014 1,014 1,014 1,014 1,014 1,014 1,014 1,014 R-squared 0.30 0.30 0.30 0.30 0.31 0.31 0.06 0.07 0.07 0.07 0.07 0.07

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level. Hours Billed is expressed in 1000 hours. New Client Revenue is expressed in 100,000 U.S dollars.

33

TABLE 8 – PERFORMANCE: CAREER ASPIRATIONS

Hours Billed New Client Rev. Hours Billed

New Client Rev.

Hours Billed

New Client Rev.

Hours Billed

New Client Rev.

IV1 IV2 IV1+IV2

[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]

Female -0.105*** -0.134*** -0.0969*** -0.147** -0.289*** -0.232* -0.122 0.0868 -0.0801* 0.103 -0.114*** -0.0836 -0.106*** -0.0364

[0.0315] [0.0332] [0.0337] [0.0706] [0.101] [0.122] [0.125] [0.261] [0.0439] [0.169] [0.0385] [0.142] [0.0376] [0.140]

Aspirations 0.0217*** 0.0181*** 0.0639*** 0.0792*** 0.0316* 0.196*** 0.0115 0.0867** 0.0166 0.114***

[0.00492] [0.00668] [0.0182] [0.0247] [0.0171] [0.0656] [0.0118] [0.0436] [0.0109] [0.0406] Female* Aspirations 0.00775 -0.0326

[0.00965] [0.0357]

Constant 1.411*** 1.903*** 0.682 0.722 0.636 0.5 0.127 0.0368 1.704*** -0.705 0.796 -0.473 1.799*** -0.191

[0.469] [0.412] [0.486] [0.489] [1.510] [1.803] [1.506] [1.510] [0.421] [1.621] [0.502] [1.854] [0.412] [1.533]

Individual Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Firm Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Region FE Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Observations 1,014 644 644 644 1,014 644 644 644 644 644 644 644 644 644

R-squared 0.3 0.265 0.288 0.289 0.066 0.029 0.049 0.05

F-test of excl. instr. 54.16*** 125.7*** 75.77*** Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level. Hours Billed is expressed in 1000 hours. New Client Revenue is expressed in 100,000 U.S. dollars. Aspirations refer to how strongly the lawyer aspires to attain equity partnership. The variable takes values 1 to 10, where 1 represents not at all and 10 Very High Aspirations. IV1 is the response to: How satisfied are you with your decision to become a lawyer? And IV2 is the response to: How much longer would you like to stay with your current employer? Both questions are asked in the first wave (2002).

34

TABLE 9 – PERFORMANCE: OVERBILLING

Hours Billed New Client Rev. [1] [2] [3] [4] [5] [6] Female -0.105*** -0.0992*** -0.111*** -0.289*** -0.283*** -0.299*** [0.0315] [0.0315] [0.0342] [0.101] [0.102] [0.111] Less than Others -0.107*** -0.137** -0.116 -0.161 [0.0397] [0.0534] [0.128] [0.173] Female* Less than Others 0.0674 0.0978 [0.0792] [0.256] Constant 0.666 0.673 1.401*** 0.165 0.174 0.175 [0.477] [0.476] [0.468] [1.536] [1.537] [1.537] Individual Controls Yes Yes Yes Yes Yes Yes Firm Controls Yes Yes Yes Yes Yes Yes Region Fixed Effects Yes Yes Yes Yes Yes Yes Observations 1014 1014 1014 1014 1014 1014 R-squared 0.30 0.31 0.31 0.07 0.07 0.07

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level. Hours Billed and Expected Hours to Bill are expressed in 1000 hours. New Client Revenue is expressed in 100,000 U.S. dollars. Less than Others takes value one if the lawyer responds that he or she is less likely to bill for actual hours worked compared to colleagues as a reason for the difficulty meeting billables, and zero otherwise.

35

TABLE 10 – PERFORMANCE: NETWORKING

Hours Billed New Client Rev. [1] [2] [3] [4] [5] [6] Female -0.105*** -0.0937*** -0.123*** -0.289*** -0.331*** -0.282** [0.0315] [0.0335] [0.0379] [0.101] [0.106] [0.120] Network Time 0.00183 -0.00218 0.0288** 0.0356** [0.00362] [0.00438] [0.0115] [0.0139] Female*Network Time 0.0126 -0.0213 [0.00778] [0.0247] Constant 0.666 1.397*** 1.412*** 0.636 0.846 0.819 [0.477] [0.478] [0.478] [1.510] [1.518] [1.518] Individual Controls Yes Yes Yes Yes Yes Yes Firm Controls Yes Yes Yes Yes Yes Yes Region Fixed Effects Yes Yes Yes Yes Yes Yes Observations 1014 939 939 1014 939 939 R-squared 0.30 0.29 0.29 0.07 0.09 0.09

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level. Network Time is the amount of time a lawyer spends attending networking functions and/or participating in recreational activities (e.g., golf) for networking purposes with other lawyers or clients.

TABLE 11 – PERFORMANCE: WORKING ON WEEKEND

Hours Billed New Client Rev. [1] [2] [3] [4] [5] [6] Female -0.105*** -0.0952*** -0.133*** -0.289*** -0.276*** -0.228* [0.0315] [0.0320] [0.0395] [0.101] [0.106] [0.131] Weekend Time 0.0144*** 0.0108** 0.0280** 0.0326** [0.00364] [0.00426] [0.0120] [0.0141] Female*Weekend Time 0.0132 -0.0169 [0.00815] [0.0269] Constant 0.666 0.627 0.628 0.165 0.292 0.29 [0.477] [0.476] [0.476] [1.536] [1.571] [1.572] Individual Controls Yes Yes Yes Yes Yes Yes Firm Controls Yes Yes Yes Yes Yes Yes Region Fixed Effects Yes Yes Yes Yes Yes Yes Observations 1014 965 965 1014 965 965 R-squared 0.30 0.30 0.30 0.07 0.08 0.08

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level. Weekend Time is the amount of time a lawyer spends working away from the office or firm on weekends.

36

TABLE 12 – PERFORMANCE: AREAS OF LAW

Hours Billed New Client Rev. [1] [2] [3] [4] Female -0.105*** -0.0846*** -0.289*** -0.310*** [0.0315] [0.0314] [0.101] [0.104] Constant 0.666 1.708*** 0.165 0.864 [0.477] [0.468] [1.536] [1.549]

Individual Controls Yes Yes Yes Yes Firm Controls Yes Yes Yes Yes Region Fixed Effects Yes Yes Yes Yes Areas of Law No Yes No Yes Observations 1,014 1,014 1,014 1,014 R-squared 0.3 0.362 0.066 0.1

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level. Hours Billed is expressed in 1000 hours. New Client Revenue is expressed in 100,000 U.S. dollars. Areas of law refers to the proportion of time devoted to each the following legal disciplines: General Practice, Antitrust, Bankruptcy, Civil Litigation, Civil Rights, Commercial Law, Criminal Law, Employment Law (Management), Employment Law (Unions), Environmental Law, Family Law, General Corporate Law, Immigration Law, Insurance, Intellectual Property, Municipal Law, Personal injury (Plaintiff), Personal Injury (Defense), Probate (Wills and Trusts), Public utilities and Administrative Law, Real Estate (Commercial), Real Estate (Personal), Securities, Tax Law, Health Law, Workers’ compensation and ‘Other’ areas.

37

TABLE 13 – EARNINGS: OVERALL

Log(annual earnings) [1] [2] [3] [4] Female -0.245*** -0.184*** -0.182*** -0.126*** [0.0226] [0.0242] [0.0234] [0.0208] Age -0.0133*** -0.0115*** -0.0026 [0.00231] [0.00227] [0.00201] Married 0.0285 0.0165 0.00841 [0.0298] [0.0292] [0.0254] No. Children 0.00623 0.00919 0.0149 [0.0135] [0.0133] [0.0115] Child Aged <4 0.0488 3.36E-02 0.0277 [0.0319] [0.0311] [0.0272] White -0.0323 -0.0208 -0.0553** [0.0272] [0.0267] [0.0234] Tenure 0.0176*** 0.0137*** 0.00823** [0.00460] [0.00452] [0.00404] Full-Time 0.653*** 0.616*** [0.0472] [0.0409] Hours 0.0153*** [0.000874] Constant 11.62*** 11.36*** 11.18*** 11.08*** [0.0153] [0.103] [0.102] [0.211] Firm Controls No No No Yes Observations 2,961 2,687 2,687 2,687 R-squared 0.05 0.15 0.18 0.38

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level.

38

TABLE 14 – EARNINGS: DIFFERENT SAMPLES

Log(annual earnings)

ALL Bill Hours Govt. & Non-Profit

Other Firms

[1] [4] [2] [3] Female -0.126*** -0.100*** -0.107*** -0.158*** [0.0208] [0.0299] [0.0313] [0.0533] Age -0.00259 -0.00498 -0.00139 0.00193 [0.00201] [0.00304] [0.00291] [0.00541] Married 0.00841 -0.0336 -0.0183 0.053 [0.0254] [0.0371] [0.0364] [0.0664] No. Children 0.0149 0.025 0.00992 0.0259 [0.0115] [0.0153] [0.0189] [0.0309] Child Aged <4 0.0277 -0.014 0.011 0.0118 [0.0272] [0.0366] [0.0432] [0.0693] White -0.0553** -0.0609* -0.118*** -0.0216 [0.0234] [0.0348] [0.0330] [0.0616] Tenure 0.00823** 0.0112** 0.00233 -0.0119 [0.00404] [0.00559] [0.00634] [0.0108] Full-Time 0.616*** 0.462*** 0.722*** 0.547*** [0.0409] [0.0594] [0.0692] [0.121] Constant 11.08*** 11.84*** 10.52*** 11.16*** [0.211] [0.480] [0.156] [0.242] Firm Controls Yes Yes Yes Yes Observations 2,687 1,124 741 478 R-squared 0.378 0.41 0.336 0.137

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level.

39

TABLE 15 – EARNINGS: INCLUDING PERFORMANCE MEASURES

Log (annual earnings) [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] Female -0.100*** -0.0990*** -0.0665** -0.0865*** -0.0554* -0.181*** -0.103*** -0.173*** -0.0943*** -0.0567* -0.0588** [0.0299] [0.0306] [0.0293] [0.0305] [0.0291] [0.0343] [0.0303] [0.0344] [0.0304] [0.0291] [0.0293] Hours Billed (’000) 0.308*** 0.304*** 0.507*** 0.508*** 0.499*** 0.581*** [0.0298] [0.0296] [0.0287] [0.0286] [0.0913] [0.207] New Client Rev 0.0433*** 0.0401*** 0.0267** 0.0290*** 0.0768*** 0.0497* (’00000 US $) [0.00967] [0.00919] [0.0118] [0.0103] [0.0185] [0.0284]

Hours Billed2 -0.0646** -0.126 [0.0285] [0.147]

New Client Rev2 -0.00240** 0.00321 [0.00110] [0.00457]

Hours Billed3 0.0134 [0.0309]

New Client Rev3 -0.000198 [0.000156] Constant 11.84*** 11.57*** 11.36*** 11.56*** 11.36*** 11.81*** 10.88*** 11.80*** 10.86*** 11.36*** 9.280*** [0.480] [0.465] [0.441] [0.460] [0.437] [0.0214] [0.0561] [0.0222] [0.0563] [0.435] [0.439] Individual Controls Yes Yes Yes Yes Yes No No No No Yes Yes Firm Controls Yes Yes Yes Yes Yes No No No No Yes Yes Region Fixed Effects No Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Observations 1,124 1,014 1,014 1,014 1,014 1,039 1,039 1,039 1,039 1,014 1,014 R-squared 0.41 0.39 0.45 0.40 0.46 0.03 0.26 0.03 0.26 0.47 0.47

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level.

40

TABLE 16 – EARNINGS: RETURNS TO PERFORMANCE

Earning per Hour

Billed Earning/New Client Rev.

[1] [2] Female -22.16 2.805 [15.58] [2.029] Age 3.973** 0.013 [1.560] [0.184] Married -9.951 -2.71 [19.37] [2.559] No. Children 11.14 -0.856 [8.069] [0.966] Child Aged <4 -24.86 1.705 [18.99] [2.439] White -7.662 -0.371 [18.65] [2.569] Tenure 0.989 -0.528 [2.909] [0.381] Full-Time -0.728 4.968 [30.60] [3.983] Constant -52.41 -3.7 [232.2] [23.67] Firm Controls Yes Yes Region Fixed Effects Yes Yes Observations 1,014 502 R-squared 0.58 0.09

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level.

41

TABLE 17 – EARNINGS: ROBUSTNESS

Log(Annual Earnings) [1] [2] [3] [4] Female -0.0990*** -0.0554* -0.0556* -0.0342 [0.0306] [0.0291] [0.0296] [0.0292] Hours Billed (’000) 0.304*** 0.304*** 0.344*** [0.0296] [0.0305] [0.0299] New Client Rev. 0.0401*** 0.0400*** 0.0394*** (’00000 US$) [0.00919] [0.00932] [0.00909] UG Uni Ranking 0.000653 [0.00389] Law School Ranking 0.0518*** [0.0144] Judicial Clerk -0.0283 [0.0819] Moot Court 0.0339 [0.0277] General Journal 0.0767** [0.0324] Specific Journal 0.0333 [0.0324] Constant 9.944*** 9.490*** 11.11*** 10.88*** [0.457] [0.431] [0.448] [0.317] Individual Controls Yes Yes Yes Yes Firm Controls Yes Yes Yes Yes Region Fixed Effects Yes Yes Yes Yes Areas of Law No No No Yes Observations 1,014 1,014 974 1,014

R-squared 0.39 0.46 0.47 0.52

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level. Areas of law refers to the proportion of time devoted to each the following legal disciplines: General Practice, Antitrust, Bankruptcy, Civil Litigation, Civil Rights, Commercial Law, Criminal Law, Employment Law (Management), Employment Law (Unions), Environmental Law, Family Law, General Corporate Law, Immigration Law, Insurance, Intellectual Property, Municipal Law, Personal injury (Plaintiff), Personal Injury (Defense), Probate (Wills and Trusts), Public utilities and Administrative Law, Real Estate (Commercial), Real Estate (Personal), Securities, Tax Law, Health Law, Workers’ compensation and ‘Other’ areas. Undergraduate GPA and Law School GPA take value 1 to 8, where 1 is the lowest and 8 is the highest. Undergraduate Uni Ranking and Law School Ranking are bracketed rankings based on U.S. News reports of 1996 and 2003 for undergraduate and law school studies, respectively. Both variables are redefined such that the higher the value, the more prestigious is the educational institution. Judicial Clerk takes value one if the lawyer has held a position as a judicial clerk in state or federal courts and zero otherwise. Moot Court takes the value one if the lawyer participated in simulated mock trials as a student and zero otherwise. General (Specific) Journal takes value one if lawyer participated in law journals’ editorial activities as a student and zero otherwise.

42

TABLE 18 – EARNINGS: ROBUSTNESS

(SELECTION INTO BILLING HOURS)

Pr(Bill Hours) [1] [2] [3] Female -0.0375*** -0.0343*** -0.0431 [0.0127] [0.0129] [0.0509] Law School Ranking 0.0205*** 0.0198*** [0.00560] [0.00701] Female*Law School Ranking 0.00182 [0.0102] Constant -0.0806 -0.177 -0.174 [0.138] [0.124] [0.126]

Individual Controls Yes Yes Yes Firm Controls Yes Yes Yes Region Fixed Effects Yes Yes Yes Observations 2,733 2,667 2,667 R-squared 0.612 0.611 0.611

Notes: * denotes significance at the 10% level,** denotes significance at the 5% and *** denotes significance at the 1% level.

43

FIGURE 1 – EVOLUTION OF LAWYERS’ GENDER GAP IN EARNINGS, 2000-2010

Note: Median weekly earnings for lawyers in the period 2000 to 2010. Population Survey’s Household Data detailed by occupation (Bureau of Labor Statistics, US).

FIGURE 2 – ASPIRATIONS TO BECOME EQUITY PARTNER

Note: Responses to the question: “How strongly do you aspire to attain the Equity Partner position within your firm?” (After the JD study, 2007).